OMB Memo Weeks Worked and Income Testing

General Information Collection Request

2017 American Community Survey Respondent Burden Testing: Cognitive Testing of the “Weeks Worked” and “Income” Series of Questions on the American Community Survey

Request: The U.S. Census Bureau, in its continuing effort to reduce respondent burden on the American Community Survey (ACS), plans to conduct additional research under the generic clearance for questionnaire pretesting research (OMB number 0607-0725). Specifically, the U.S. Census Bureau will cognitively test proposed changes to the “Weeks Worked” and the “Income” series of questions on the ACS. For the “2017 ACS Respondent Burden Testing,” the U.S. Census Bureau has contracted with Westat, a statistical survey research corporation headquartered in Rockville, Maryland, to perform cognitive testing on the proposed changes.

Purpose: The revised questions, developed under the auspices of the OMB Interagency Group on the American Community Survey and related subject-area subcommittees, are designed to facilitate respondent comprehension, reduce respondent burden, and where possible, streamline wording.

For the “Income” question, the current version of the ACS asks respondents to provide between 8 and 10 components of income (recipiency (i.e., source) and amount), the sum of which provides the respondent’s “total income” for the past 12 months. The number of components asked depends upon the mode of data collection and previous survey responses. The ACS asks respondents for the individual components of income such as wages, self-employment income, and supplemental security income to ensure that respondents provide all sources of income necessary for them to derive the total income component. A number of respondents report that answering these income questions is difficult, burdensome, and sensitive. For these reasons, we are exploring using administrative records data to replace or supplement income questions on the ACS.

One idea being pursued is to restructure the income questions to only collect recipiency of income (by source) as well as total income and then use administrative records data and modeling to provide the full level of information required. Therefore, we plan to cognitively test a version of the “Income” question that asks the recipiency questions, but does not ask the respondent for specific amounts, while still collecting total income. We would like to (1) study if respondents include all of the income that they replied “yes” to in their calculation of total income; (2) probe respondents about whether they could have given specific amounts for each of the income sources and if they would have used references (such as tax forms or pay stubs) to look up their income information; and (3) evaluate whether respondents read and understand the changed wording and placement of instructions.

Additionally, we will test changing the reference period from “the past 12 months” to “last year”. The purpose of this change is to align the ACS reference date to the administrative records data. Moreover, we believe that respondents may be able to provide total income easier as this approach aligns with tax reporting as well. However, this may be affected by which part of the year the interview is being conducted. Therefore, this cognitive testing will be conducted at the end of the calendar year as well as the beginning of the calendar year to reflect potential extremes in reporting issues. We would like to know (1) if respondents can comprehend the reference period; (2) if respondents who report at the end of the year can accurately report income for the previous calendar year; (3) if respondents reporting at the beginning of the year before they do their taxes can report for the previous calendar year; and (4) if the revised reference period affects respondent burden (Is reporting income for the last full calendar year easier than income for the past 12 months?)

The “Weeks Worked” series of questions on the current version of the ACS that we will investigate is comprised of three work-related questions that asks (paraphrased): how many weeks during the past 12 months did the respondent work (i.e., “Weeks Worked”); how many hours did the respondent work in the “Weeks Worked” (i.e., “Hours Worked”). For the “Weeks Worked” series of questions, we will investigate and cognitively test the following proposed changes:

• Sequence of “Hours Worked” and “Weeks Worked”: Evaluate asking “Hours Worked” before “Weeks Worked” to determine if asking “Hours Worked” first provides helpful context in framing the work done in the past year.

• Months Worked: Evaluate the explicit option to respond to “Weeks Worked” in months in the interviewer-administered mode. We would like to know (1) if respondents choose this option; (2) if this option lessens burden; and (3) if respondents report accurately when reporting this way (we are concerned that respondents who work 2 weeks a month will incorrectly report 12 months of work instead of 24 weeks.

• Reference Period: Examine modifications to the reference period to align with the proposed changes to the reference period for the “Income” questions. Specifically, we would like to know: (1) if respondents comprehend the revised reference period; (2) if respondents report accurately based on this reference period; and (3) if the revised reference period affects respondent burden.

Population of Interest: The planned cognitive testing will focus on assessing and improving the experience for the general population.

Timeline: Cognitive interviewing will take place between December 1, 2016 and February 28, 2017, with recruiting beginning in late November. There will be two rounds of cognitive testing: the first round will take place in December of 2016 using a “2015” reference period. The second round will take place in January and February of 2017 using a “2016” reference period. Westat plans to conduct the interviews at its Rockville, Maryland location. If necessary, Westat will use additional locations to facilitate recruitment.

Language: Testing will be conducted in English only.

Method: For each round of testing, Westat will perform 24 English-language cognitive interviews –– 48 cognitive interviews in total. In each round, we will conduct both paper questionnaire interviews (self-administered) and interviewer-administered interviews. For each interview, a respondent will answer a set of questions specific to his or her assigned mode. As part of the cognitive interview process, Westat interviewers will probe the respondents’ understanding and perception of the questions. Each cognitive interview will last approximately 40 minutes. The interviews will be audiotaped and may be subject to one-way observation from designated sworn employees of Westat and the U.S. Census Bureau.

Sample: Westat will screen approximately 164 people to obtain 48 completed interviews.

Recruiting will take place between November 28, 2016 and February 10, 2017. All respondents must reside in households with at least one other adult. We will also strive for each respondent to meet the requirements for both survey topics (see below) and will try to get as many respondents as possible who have multiple sources of income.

• Weeks Worked: We will include respondents who worked “irregular” job schedules in 2015 or 2016 (i.e., less than 52 weeks per year, do not work each week or the same hours each week, have seasonal pattern work, etc.), worked “regular” jobs with regular schedules in 2015 or 2016, worked in 2016 but not in 2015, and have not worked at all in the past five years. More respondents will be recruited that had irregular work schedules and regular work schedules than those who did not work at all.

• Income: We will seek respondents with the following characteristics:

• Respondents who do not work or are retired (earning a pension or 401k) along with those who have not worked in the past five years (in recognition that we need mostly workers for weeks worked, this will be a smaller number of respondents).

• Respondents who are currently working (employed with a wage/salary job)

• Respondents who receive these income types:

• Self-employment income

• Interest or dividend

• Net rental income

• Commission or tips

• Social security

• Supplemental security income

• Public assistance income

• Respondents who are married and receive self-employment income from a business or interest and dividends, or net rental income (we want to know how they report joint income)

Recruitment: To recruit 48 respondents that meet the specified characteristics, Westat will use its own participant database as a starting point, and supplement it with recruits obtained through print advertising. See the attached OMB-approved flyer template and the Craigslist ad template (Attachment B). If necessary, Westat will also use make use of its personal and business networks, and target commercial locations that attract the desired respondents. Westat will use a screener (see Attachment B) to screen potential respondents who respond to any of its outreach efforts.

Protocol: The testing protocols by mode are found in Attachments C and D. The protocols may be updated between the December and January interviewing. The cognitive test interviewer-administered survey questions are found in Attachment E, and the paper survey instrument is found in Attachment G.

Use of Incentive: Respondents who are selected and complete an interview will receive $40 to offset the cost of participation (e.g., transportation, childcare costs).

Materials to be Used for this Project:

A: ACS Cognitive Testing to Reduce the Burden and Difficulty of Questions: Research Goals, Recruiting Requirements, and Question Wording for the Second Round of Testing

B: ACS Respondent Burden Testing – Recruiting Screener and Advertisements

C: ACS Cognitive Testing Protocol (Paper)

D: ACS Cognitive Testing Protocol (CAI)

E: Interviewer-Administered Survey Questions

F: Interviewer-Administered Roster Questions

G: Paper Survey Questions

Length of interview: Screening will take approximately 5 minutes per person. Therefore, the initial screening of 164 people will take approximately 14 hours. An interview will take approximately 40 minutes per respondent. Therefore, the total number of hours to interview 48 respondents will take 32 hours. The maximum burden is approximately 46 hours.

The contact person for questions regarding data collection and study design is:

Agnes Kee

American Community Survey Office

Room # 4K271

Washington, DC 20233

(301) 763-1516

Agnes.S.Kee@census.gov